CORE Metadata, citation and similar papers at core.ac.uk
Provided by Stellenbosch University SUNScholar Repository
Typology of Smallholder Farming in South Africa’s Former Homelands: Towards an Appropriate Classification System
By
Petrus Louw Pienaar
Thesis presented in partial fulfilment of the requirements for
the degree of Master of Science in Agriculture (Agricultural
Economics) in the Faculty of AgriSciences at Stellenbosch
University
Supervisor: Lulama Traub
December 2013 Stellenbosch University http://scholar.sun.ac.za
Declaration
By submitting this thesis electronically, I declare that the entirety of the work contained therein is my own, original work, that I am the sole author thereof (save to the extent explicitly otherwise stated), that reproduction and publication thereof by Stellenbosch University will not infringe any third party rights and that I have not previously in its entirety or in part submitted it for obtaining any qualification.
Date: ……………………………….. 2013/11/25
Copyright © 2013 Stellenbosch University
All rights reserved
i
Stellenbosch University http://scholar.sun.ac.za
Dedication
I dedicate this master’s thesis to my Heavenly Father for giving me both the opportunity and desire to complete this research. To Him belong all the glory and honour in Christ Jesus. He has transformed my life and has given me a hope and a future. Without Him, I am nothing.
Acknowledgements
Then I would also like to express my gratitude to my supervisor, Ms Lulama Traub, for the time and effort she put into this thesis. You have been an inspiration and a source of motivation. To my family and friends, thank you all for the support and the love I have received from you.
ii
Stellenbosch University http://scholar.sun.ac.za
Abstract
Typology of Smallholder Farming in South Africa’s Former Homeland: Towards an Appropriate Classification System
PL Pienaar
Department of Agricultural Economics, University of Stellenbosch, Private Bag X1, Matieland 7602, South Africa
Thesis: MScAgric December 2013
The agriculture sector continues to be viewed as a vehicle through which economic growth and development can be achieved; particularly for developing economies. This view is incorporated in South Africa’s rural development framework in the National Development Plan, which indicated that this sector will be the main driver in developing the country’s rural economies. However, the South African agricultural sector is known to be dualistic; consisting of a large-scale commercial and a small-scale subsistence sector. This study is particularly focused on smallholder farming in South Africa, which have developed as a result of the decades of government intervention that have guided reform driven by the general political and economic philosophy of white domination. The most notable interventions, which drew the line between white and black landholding, were the Natives’ Land Acts of 1913 and 1936, followed by various policy interventions to support White, large-scale agriculture.
The question remains whether or not an expanded smallholder sector can significantly contribute to rural development, employment creation and poverty reduction in the former homeland areas of South Africa. In order to answer this question, the need arises for reliable data on smallholder farming, conceptual clarification on definitions of “smallholder” or “small-scale” farmers and diversity among farming systems needs to be taken into account. These considerations are crucial in order to design and implement effective rural development policies.
iii
Stellenbosch University http://scholar.sun.ac.za
One way of addressing this question is the use of farm typologies. Given the diversity that exists within agricultural systems, various schemes of classification have been developed and evolved over time. The objective of this study is to provide an empirical framework that would classify smallholder farmers in the former homeland areas of South Africa according to their livelihood strategies. This study seeks to achieve the objective in three distinct ways. Firstly, by giving a broad overview of the smallholder sector in South Africa. Secondly, by utilizing Geographic Information Systems (GIS) techniques to identify farming households situated in the former homeland areas, using the General Household Survey (GHS) and the Income and Expenditure Survey (IES). Thirdly, apply multivariate statistical techniques, specifically Principle Component Analysis (PCA) and Cluster Analysis (CA), to develop the ultimate classification system.
The results from both typologies suggested eight distinct types or groups of farming households in the former homeland areas. Important findings suggest that higher salary incomes are crucial for the enablement of households to market their produce. Social grants were found to be key in determining livelihood strategies among faming households, most notably old age and child support grants. One of the groups that were identified was typically food insecure, with their agricultural production not sufficiently feeding the household. Lastly, direct agricultural support from the government was clearly focused on livestock services which placed a minority of households at a distinct advantage to sell produce to the market.
iv
Stellenbosch University http://scholar.sun.ac.za
Uittreksel
‘n Tipologie van Kleinskaalse Boerdery in Suid-Afrika se Voormalige Tuislande: Die Ontwikkeling van ‘n Toepaslike Klassifikasiesisteem
PL Pienaar
Departement van Landbou Ekonomie, Universiteit van Stellenbosch, Privaat Sak X1, Matieland 7602, Suid-Afrika
Tesis: MScAgric Desember 2013
Die landbousektor word algemeen gesien as een van die moontlike drywers vir ekonomiese groei en landelike ontwikkeling, spesifiek in ontwikkelende lande. Hierdie siening word ook uitgesonder deur die Suid-Afrikaanse ontwikkelingsraamwerk, en by name in die Nasionale Ontwikkelingsplan wat aandui dat die landbousektor die hoofrol behoort te vervul om landelike gebiede te ontwikkel. Die vermoë om hierdie mandaat uit te voer moet in die konteks van die kenmerkende dualisme raakgesien word. Suid-Afrika het hoofsaaklik twee tipes boere; grootskaalse kommersiële boere en kleinskaalse, meestal bestaansboere, wat meestal in die voormalige tuislande opereer. Hierdie dualisme is die resultaat van verskeie regeringsinmengings, hoofsaaklik gedryf deur die algemene politieke bestel, ideologie en beleid wat op rasseklassifikasie gegrond was gedurende die vorige eeu. Sekerlik een van die mees bekende was die Naturellegrond Wet van 1913 en 1936, wat die skeidingslyn tussen swart en wit grondbesit ingestel het. Verder is verskeie wetgewings implimenteer om die kommersiële landbousektor te bevoordeel gedurende hierdie tydperk..
In hierdie konteks is dit belangrik om te vra of die uitbreiding van die kleinskaalse landbousektor werklik kan bydra tot landelike ontwikkeling, werkskepping en armoedeverligting in die voormalige tuislande van Suid-Afrika. Om hierdie vraag te beantwoord word betroubare inligting benodig, moet die konsep van “kleinskaalse boere” uitgeklaar word en laastens moet diversiteit tussen verskillende boerderystelsels in ag geneem
v
Stellenbosch University http://scholar.sun.ac.za
word. Die antwoorde op hierdie vrae is noodsaaklik vir die ontwikkeling en implimentering van effektiewe landelike ontwikkelingsbeleid.
Die gebruik van boerderytipologieë is ‘n oplossing om hierdie kwessies aan te spreek. Verskeie klassifikasiesisteme is in die verlede ontwikkel om die diversiteit in boerderystelsels te ondersoek. Die hoof doel van hierdie studie is om ‘n empiriese raamwerk te ontwikkel om kleinskaalse boerderye, wat in die voormalige tuislande voorkom, volgens hul lewensbestaanstrategieë te klassifiseer. Om hierdie doelwit te bereik, sal die studie eerstens ‘n oorsig gee van die kleinskaalse landbousektor in Suid-Afrika. Tweedens sal Geografiese Inligtingstelsels (GIS) tegnieke gebruik word om spesifiek huishoudings in die voormalige tuislande te indentifiseer in die Algemene Huishoudings Opname (AHO) en die Inkomste en Uitgawes Opname (IUO). Derdens sal meerveranderlike statistieke gebruik word, spesifiek Hoofkomponentanalise (HKA) en Bondelontleding (BO), om die klassifikasiesisteem te ontwikkel.
Die resultate van die tipologieë wat in hierdie studie ontwikkel is gee agt spesifieke groepe van boerderyhuishoudings. Hierdie groepe was beduidend verskillend van mekaar en elkeen se lewenbestaanstrategieë word uitgewys. Die hoofbevindings dui aan dat addisionele salarisinkomste ‘n belangrike rol speel in die vermoë van kleinskaalse boere om hul produkte te verkoop. Verder is dit opmerklik dat maatskaplike toelaes ‘n aansienlike rol gespeel het in die vorming van die groepe, spesifiek wat betref ouderdomspensioene en kindertoelae. Daar is ook ‘n spesifieke groep huishoudings in beide tipologieë wat probleme ondervind om voedselsekuriteit op huishoudelike vlak te handhaaf. Laastens wys die studie dat direkte landbou-ondersteuning teenoor kleinskaalse boere ‘n kenmerkende fokus op lewendehaweboerderye plaas wat sulke boerderye bevoordeel het om vir die mark te produseer.
vi
Stellenbosch University http://scholar.sun.ac.za
Contents
Declaration………………………………………………………………………………….....i
Dedication…………………………………………………………………………….………ii
Acknowledgements………………………………………………………………….…....…..ii
Abstract………………………………………………………………………………………iii
Uittreksel…..…………………………………………………………..……………………..v
Contents……………………………………………………………………………..…….....vii
List of Figures...…………………………………….………………………………...……...ix
List of Tables…………………………………………………………………………..….....ix
List of Maps…………………………………………………………………………………..x
1 Introduction ...... 1 1.1 Objectives of this study ...... 3 1.2 Study Outline...... 4 2 An Overview of the South African Smallholder Sector ...... 5 2.1 Introduction ...... 5 2.2 The Dualistic Agricultural System in South Africa ...... 5 2.2.1 Post-Apartheid: 1994 - 2013 ...... 7 2.3 The Agrarian Structure in the 21 century ...... 8 2.4 Defining Smallholder Farming Systems ...... 9 2.4.1 Global Definition ...... 10 2.4.2 Smallholder Farming within South Africa ...... 11 2.5 Smallholder livelihoods in South Africa ...... 13 2.6 Support to smallholder farming in South Africa ...... 14 2.7 Conclusion ...... 16 3 Literature Review on Typology Development ...... 18 3.1 Classification within Agriculture ...... 18 3.1.1 Farming Styles ...... 19 3.1.2 Activity Systems ...... 20
vii
Stellenbosch University http://scholar.sun.ac.za
3.2 Defining Farm Typologies ...... 21 3.3 Rationale for developing smallholder Typologies ...... 22 3.4 Theoretical Background for Creating Typologies ...... 22 3.5 Approaches to Constructing Typologies ...... 24 3.5.1 Qualitative Approach ...... 25 3.5.2 Quantitative Approach ...... 27 3.6 Conclusion ...... 30 4 Methodology ...... 33 4.1 Introduction ...... 33 4.2 Methods ...... 33 4.2.1 Step 1: Theoretical Framework ...... 33 4.2.2 Step 2: Data and study area ...... 34 4.2.3 Step 3: Variable selection ...... 38 4.2.4 Step 4: Principle Component Analysis ...... 42 4.2.5 Step 5: Cluster Analysis ...... 44 4.2.6 Step 6: Cluster Validation - ANOVA ...... 47 4.3 Conclusion ...... 48 5 Results and Classification of Smallholder Farming Households ...... 50 5.1 Introduction ...... 50 5.2 Descriptive Analysis ...... 50 5.2.1 Household characteristics ...... 51 5.2.2 Income and Expenditure ...... 55 5.2.3 Agricultural Orientation ...... 56 5.3 Results ...... 61 5.3.1 Principle Component Analysis Results ...... 61 5.3.2 GHS Typology 1 ...... 62 5.3.3 IES Typology 1 ...... 66 5.4 Cluster Analysis Results ...... 68 5.4.1 GHS Typology 1 ...... 68 5.4.2 IES Typology 1 ...... 73 5.3 Discussions and Conclusion ...... 77 5.3.1 Main findings ...... 77 5.3.2 Results compared to other South African smallholder typologies ...... 81
viii
Stellenbosch University http://scholar.sun.ac.za
6 Conclusions and Recommendations ...... 83 6.1 Thesis Overview ...... 83 6.2 Key Findings ...... 85 6.3 Policy recommendations ...... 87 7 List of References ...... 89 8 Appendix ...... 104 8.1 GHS Typology 1 ...... 104 8.2 IES Typology 1 ...... 108
List of Figures
Figure 1: Graphical representation of the terminology used to define smallholder farmers in South Africa ...... 13 Figure 2: Steps used in both qualitative and quantitative typology development ...... 27 Figure 3: Age distribution and employment numbers of the former homeland farming population ...... 52 Figure 4: Age distribution of household heads within agricultural households ...... 53 Figure 5: Level of education across age groups of household heads ...... 54 Figure 6: Household monthly income distribution according to 10 income groups ...... 56 Figure 7: Livestock ownership across 10 income groups ...... 57 Figure 8: Dendogram of GHS Typology 1 showing the 8-cluster solution of farming households...... 68 Figure 9: Dendogram of IES Typology 1 showing the 8-cluster solution of farming households...... 73 Figure 10: Screeplot from PCA results for GHS Typology 1 ...... 104 Figure 11: Screeplot from PCA results of IES Typology 1 ...... 108
List of Tables
Table 1: Number of black farming households in South Africa ...... 12 Table 2: Government agricultural support to black farming households ...... 16 Table 3: Land area of the Former Homelands ...... 36 Table 4: Sample size of smallholder households ...... 38
ix
Stellenbosch University http://scholar.sun.ac.za
Table 5: Variables used in the typology of smallholder farming households ...... 40 Table 6: Descriptive statistics ...... 51 Table 7: Distribution of land sizes of crop producing households ...... 57 Table 8: Percentage share of farm expenditure categories of household farm inputs per annum ...... 58 Table 9: Government agricultural support to farming households ...... 59 Table 10: Key variables of farming households ...... 60 Table 11: Tenure arrangement specifications ...... 61 Table 12: Validation for Factor Analysis ...... 61 Table 13: Output of factor loadings from PCA for GHS Typology 1 ...... 63 Table 14: Output of factor loadings from PCA for IES Typology 1 ...... 66 Table 15: Cluster results for GHS Typology 1 with mean values of each variable ...... 70 Table 16: Cluster results for IES Typology 1 with mean values of each variable ...... 75 Table 17: Total variance explained with rotation included in GHS Typology 1 ...... 105 Table 18: Summarized table of z-scores from PCA used for CA in GHS Typology 1 ...... 106 Table 19: Correlation matrix of the 25 included GHS Typology 1 variables in PCA ...... 107 Table 20: Total variance explained with rotation included for IES Typology 1 ...... 109 Table 21: Summarized table of z-scores from PCA used for CA in IES Typology 1 ...... 110 Table 22: Correlation matrix of the 16 included IES Typology 1 variables in PCA ...... 111
List of Maps
Map 1: Location of homelands and enumerator areas in GHS 2010 and IES 2010/11 ...... 37
x
Stellenbosch University http://scholar.sun.ac.za
Chapter 1 Introduction
The debate on the role of agriculture in development was initiated by the classical theorists, led by Arthur Lewis (1954), who viewed economic development as a growth process of relocating productive factors away from agriculture towards a modern industrial sector (Byerlee, et al., 2009). Beginning in the 1960’s, Johnston and Mellor (1961) with their seminal work, argued that agriculture was essential for growth especially in the early stages of industrialization. Following the Green Revolution in Asia, where the positive impact of agricultural growth on rural development was found to be the strongest in countries with agriculture sectors dominated by smallholder farmers, a renewed emphasis was placed on broad-based agricultural growth and productivity increases in rural economies (Mellor, 1976; Rosegrant & Hazell, 2001; Lipton, 2005; Byerlee et al., 2009; Diao et al., 2010). To date, the agriculture sector continues to be viewed as a vehicle through which economic growth and development can be achieved, particularly for developing economies where the agricultural sector is dominated by largely informal, small-scale producers (Machethe, 2004; Dercon, 2009; Christiaensen et al., 2010; De Janvry, 2010; De Janvry & Sadoulet, 2010).
In contrast, the agricultural sector in South Africa is dualistic; consisting of a large-scale commercial and small-scale subsistence sector. The commercial sector includes approximately 40 000 large-scale farming enterprises whereas the small-scale sector consists of approximately 2 million, largely subsistence, farming households (Aliber & Hart, 2009; Aliber & Cousins, 2013). Prior to 1994, policy emphasis was placed on the development and support of the formal commercial agricultural sector to the exclusion of a much larger number of smallholder, black farms located in the homeland areas (Modiselle, 2001). The historical development of the dualistic agricultural system reveals a long history of biased and distortionary policy interventions, spesifically in the colonial era (Thirtle, et al., 2000). The most notable intervention, which the drew the line between white and black landholding, was the Natives’ Land Act of 1913 and 1936. These legislative measures effectively destroyed a once thriving African farming sector which was able to compete with white settler farms during the mid-19th century (Vink & van Zyl, 1998). Furthermore, various policy instruments were introduced from 1910 onwards to support the white commercial
1
Stellenbosch University http://scholar.sun.ac.za
farming sector. These measures broadened the dualistic gap between smallholder and large- scale farming in South Africa, which are still present today.
In 1994, with the transition to democracy, agricultural policies were aimed at supporting smallholder agriculture in South Africa in the form of infrastructure grants, production inputs support, access to loans and extension services. Through such support programs, the goal of the new administration was to create a new unified economy where both large and small farm enterprises could compete in both domestic and international markets (Van Averbeke & Mohamed, 2006). However, evidence suggests that these programs have been ineffective in stimulating rural growth and poverty alleviation and the dualistic nature of the agricultural sector continues to persist; with smallholder farmers in South Africa facing challenges of limited access to markets, inputs and credit as well as constrained property rights and relatively high transaction costs (Perret et al., 2005; Ortmann & King, 2006; Hall & Aliber, 2010).
To date, the National Development Plan (2011) recognizes the importance of the agricultural sector in developing the county’s rural economies and generating employment through the creation of at least one million new jobs. The NDP also makes specific reference to the former homeland areas:
“Underdevelopment in the former homelands must be confronted through agricultural development, improved land management, infrastructure and targeted support to rural women.” (NPC, 2011).
However, the question remains as to whether or not an expanded smallholder sector can make a significant contribution to rural development, employment creation and poverty reduction in South Africa’s rural areas (Cousins, 2013). To address this question, two key problems hindering policy formulation towards smallholder farming in South Africa need to be tackled. The first is the lack of reliable and detailed empirical data on small-scale farmers. The number of survey instruments seeking detailed information regarding smallholder agriculture in South Africa has been limited and outdated (Aliber & Hart, 2009; Aliber & Hall, 2010). This apparent lack of detailed information on smallholder agriculture makes it almost impossible to understand the diversity within these agricultural production systems, each having a unique set of distinctive limitations and constraints and faces a heterogeneous decision-making environment (Van der Ploeg et al., 2009).
2
Stellenbosch University http://scholar.sun.ac.za
The second is conceptual; in other words, what exactly is meant by the term “smallholder” and “small-scale farmer”? The definition of smallholder farmers has been ambiguous and different terminology has been used to classify them (Ortmann & King, 2006). Diversity in farming systems occur within a diverse biophysical and socio-economic environment and using terms such as “smallholder” tends to obscure inequalities and class-based differences within a large group of heterogeneous farmers (Cousins, 2010; Tittonell, et al., 2010). These considerations are crucial and needs to be addressed if a farmer-focussed and farming- systems research approaches are utilized in order to designing and implement effective rural development policies (Laurent et al., 1999; Cousins, 2013).
One way of addressing the abovementioned challenges to support policy formulation is the use of farm typologies (Capillon, 1993). Typologies have a long lineage in sociology with the primary aim of distinguishing the social and economic characteristics of rural households (Whatmore, 1994; Emtage, 2004). Within the framework of rural development, a typology is a procedure (qualitative and/or quantitative) for developing and describing relatively homogenous groups of households with similar constraints and objectives, which are expected to respond to external influences in a similar fashion (Perret & Kirsten, 2000; Tefera, et al., 2004). 1.1 Objectives of this study
The goal of this study is to develop a typology of the smallholder farming sector within South Africa; particularly within the former homeland areas. This typology will provide a valuable understanding of the factors affecting farming systems in these areas and would indicate how this sector could possibly contribute to the mandate of development to create employment and reduce poverty within South Africa’s rural economies.
To achieve this goal the study objective is to provide an empirical framework that would classify smallholder farmers in the former homeland areas of South Africa by making use of sound statistical procedures. The study draws on the work done by Perret and Kirsten (2000), Anseeuw et al. (2001) and Perret et al. (2005) in using a livelihood approach to study diversity and livelihood strategies of rural households. To address this objective this study will;
1. Provide an overview of the smallholder farming sector in South Africa from the late 19th to the early 21st century.
3
Stellenbosch University http://scholar.sun.ac.za
2. Utilize Geographic Information Systems (GIS) techniques to identify farming households situated in the former homeland areas. Applying GIS techniques to sample farming households in the former homeland areas, will give a much needed description of the sector. 3. Develop a classification system of smallholder farmers utilizing Principle Component Analysis (PCA) and Cluster Analysis (CA). Using these multivariate statistical techniques will enable the development of an adequate classification system based on a range of household indicators, rather than one indicator, such as farm size. The analysis will be done by treating the household as the primary economic unit and a sustainable livelihood approach will be used to create typologies of smallholder farming households utilizing data from both the General Household Survey (GHS) and the Income and Expenditure Survey (IES). To test the robustness and validity of the proposed typologies, empirical testing in the form of an Analysis of Variance (ANOVA) is conducted to see if there are significant differences between the groups and whether the proposed typological groupings are valid and stable.
1.2 Study Outline
The first chapter gives an overview of the South African smallholder farming sector with specific reference to the history, agrarian structure and the definition of smallholder farmers. It contextualises the current environment of the sector and provides important emphasis on past policies.
Chapter Three will follow with a literature review on typology development. This will be done by introducing classification of agricultural systems and by looking at the rationale, theoretical background and approaches used to develop typologies. The different methods used to create classification systems will then be discussed, explaining the differences between qualitative and quantitative classification systems and will seek to introduce the various instruments to be used in the next chapter. Chapter Four is the chapter on the methodology used in the study and therefore explains all the steps used in the analysis to create the proposed typology of farming households in the former homeland areas of South Africa.
Chapter Five gives the results and will explain the findings on the specific group classifications. Chapter Six contains a summary of the results of the study, provides a synthesis of those results and gives some policy recommendations.
4
Stellenbosch University http://scholar.sun.ac.za
Chapter
2 An Overview of the South African Smallholder Sector
2.1 Introduction
This chapter describes the smallholder agricultural sector in South Africa within its specific environment. The elements in this chapter are important for understanding the dynamics of smallholder livelihoods in the South African agricultural economy. The chapter will start with an overview of the dualism within the sector, highlighting the important policy interventions that have caused it. This will be followed by the evaluation of the current agrarian structure in South Africa, indicating where the smallholder sector fits into the national economy. Then emphasis will turn to the definition of smallholder farmers both internationally and domestically. Finally, the rural livelihoods in South Africa will be reviewed with emphasis on farming as a livelihood strategy. 2.2 The Dualistic Agricultural System in South Africa
The current South African agriculture sector is a result of many factors in the past. The sector has played a prominent role in the economic development of South Africa, but is also well known for its lack in making full use of the available resources at its disposal. The development of the agricultural sector during the last century is characterized by structural change and the effect of political, economic, social, and historical factors that have caused the duality within the sector (Essa & Nieuwoudt, 2003). It is dualistic in the sense that it consists of a well-integrated, highly capitalized commercial sector on the one hand and a fluctuating subsistence or smallholder sector on the other (Vink & Kirsten, 2003; Aliber & Hart, 2009; May & Carter, 2009). This implies, therefore, that the South African agriculture sector has mainly two types of farmers; subsistence farmers in the former homeland areas and large- scale commercial farmers on privately owned land (Kirsten & van Zyl, 1998). Also in between these two main categories are farmers moving from subsistence agricultural production towards commercialization mostly with the help of government programmes aimed at establishing more black commercial farmers in South Africa. 5
Stellenbosch University http://scholar.sun.ac.za
The dualistic nature and division between the commercial, large-scale farming sector and the comparatively low productive, struggling smallholder sector is not merely a result of economies of scale. This distinctive, dualistic structure in the farming sector has been created as a result of the interplay between public policies and the functioning of land, labour, and capital markets over time (Vink & van Zyl, 1998). Arguably, the most detrimental discriminatory legislation towards the marginalisation of black farmers was the Natives’ Land Acts of 1913 and 1936. These legislative measures of discrimination against black farmers were instrumental in creating the dualism that currently persists within the agricultural sector (Kirsten et al., 1998).
The history of smallholder farming in South Africa is well documented in the literature (Bundy, 1979; Kirsten, et al., 1998; Thompson, 2000; Hamilton, et al., 2012). African farming societies existed many years before the arrival of white European settlers in the 17th century (Thompson, 2000; Parkington & Hall, 2009). These indigenous African societies in Southern Africa consisted of hunter-gatherers (San), pastoralists (Koikoi), and the Bantu- speaking mixed farmers (Africans) (Thompson, 2000). The arrival of European settlers, particularly those interested in farming, resulted in a long history of conflict on land acquisition between white settlers and the indigenous African tribes in South Africa (Thompson, 1995; Terreblanche, 1998; Tihanyi & Robinson, 2011). The story of disempowerment of these African farmers continued throughout the centuries leading up to the establishment of the Union in 1910. European settlers influenced the government in applying restrictive measures on black African land rights in the period leading to the establishment of the Union (Binswanger & Deininger, 1993). These policies was ushered in to suppress and isolate African farmers from mainstream agriculture, and aimed at transforming them into wage labourers (Vink & van Zyl, 1998).
Before the establishment of the Union, African farming was relatively viable during the second half of the 19th century. African farmers at the time, whether farming on private land or as tenants, proved to be as efficient as large-scale settler farmers. According to Bundy (1979) these African farmers supplied the major towns of the colony with grain and exported surplus to the Cape between 1850 and 1870. Compared to large-scale farming, these African family farming units were efficient and viable to produce agricultural products with simple technologies and plentiful land. Labour was said to be the most important success factor of farming at the time and white settlers with low profitability could not offer wages that would attract indigenous labourers, who had no need to work for wages (Mbongwa et al., 2000).
6
Stellenbosch University http://scholar.sun.ac.za
In the early part of the 20th century at the time of the establishment of the Union in 1910, the existing racial discrimination in access to land was consolidated (Vink & van Zyl, 1998). The Natives’ Land Act of 1913 drew the line between white and black landholding and prohibited any transactions for the purchase, hire or acquisition of land to black people (Mbongwa et al., 2000). Under the terms of the Land Act of 1913, 7% of the land and later 13% under the 1936 Land Act was reserved exclusively to Africans. This Act attempted to outlaw access to land such as tenancy and sharecropping, and caused much disruption to black farming production (Vink & van Zyl, 1998). The combined effects of the Land Acts and several other interventions stripped the African household farming sector of its independence and these farmers were condemned to agricultural production inside the reserves on small areas of communal land. These famers were choked of opportunities outside of the labour market and gradually capital, wealth, farming skills and information build up by centuries was being destroyed (Mbongwa et al., 2000). It is estimated that 13 million of the 40 million South Africans lived in the homeland areas at the time of transition and that 80% of the rural people were living in poverty (Lyne & Darroch, 2003).
The settler state did not only introduce the above mentioned discriminatory policies that crippled the African farming sector, but also introduced a wide range of instruments to support white commercial farming (Vink & van Zyl, 1998). The main objective was to use factors of production optimally with respect to development and the state supported white farmers through legislation such as the Cooperative Societies Acts and the Marketing Acts, through investment in research and development, infrastructure and extension services (Vink, 2009). Subsequent measures of protection that followed to achieve these objectives for “white Agriculture” were administered prices, input subsidization, import controls, compulsory single-channel marketing and export subsidies and generous disaster assistance towards agriculture (Oettle et al., 1998; Makhura & Mokoena, 2003).
2.2.1 Post-Apartheid: 1994 - 2013 The transition from Apartheid to a new democratically elected government in 1994 brought about various policy changes to transform the agricultural sector to an open economy. Policy changes in the agricultural sector included the deregulation of the marketing system, abolition of certain tax concessions, reduction in expenditure from the national budget, land reform, trade reform and new labour legislation (Groenewald & Nieuwoudt, 2003). With these changes in policy many African farmers expected positive changes to the agricultural sector.
7
Stellenbosch University http://scholar.sun.ac.za
The African National Congress (ANC) stipulated that the improvement of small-scale agricultural production and increased participation of emerging farmers in the economy were the pillars of the Reconstruction and Development Programme (RDP) (Makhura & Mokoena, 2003). The general aim of the new agricultural policy was to create a new unified economy where both large and small farm enterprises could compete in harmony in the domestic and international markets (Van Averbeke & Mohamed, 2006). The ruling party indicated that a larger and more vibrant small-scale farming sector had the potential to address key issues such as rural poverty, unemployment and food insecurity (Aliber & Hall, 2010).
Towards undoing the effects of decades of policies that affected black South Africans, the new government initiated a series of land reform programmes from 1994 with the intent to redistribute 30% of the white owned land to previously disadvantaged people. The intention was to make land accessible, to enable security of tenure for these rural people, and to improve small-scale production capacity. Three main land reform instruments were used: Land restitution, Land tenure reform and Land redistribution (Lyne & Darroch, 2003). 2.3 The Agrarian Structure in the 21 century
The agricultural sector continues to be characterized by inequality in terms of the distribution of economic assets, support services, market access, infrastructure and income (Oettle et al., 1998). The total land area in South Africa is approximately 122.3 million hectares of which farmland consists of 100.6 million. Among the 100.6 million hectares of farmland, 83.3% is grazing land and only 16.7 million hectares are considered potential arable land (DBSA, 2000). Furthermore, only 1.35 million hectares of the potential arable land in South Africa is available for irrigation and accounts for more than one third of the total output in the agriculture sector (Vink & Kirsten, 2003). Commercial agriculture takes up 86.2 million hectares of the available farmland, while the smallholder sector only utilizes 14.5 million hectares (Fenyes & Meyer, 2003; DAFF, 2012a).
Commercial agriculture in South Africa is made up of a relatively small number of commercial farmers who occupy 87% of total agricultural land and are responsible for over 95% of agricultural production in South Africa (Vink & Kirsten, 2003). The number of farmers have been declining over time from 60 000 in 1996 to approximately 45 000 by 2002. More recently, this number was estimated to be even lower at 35 000 large-scale, mostly white, farmers (Aliber & Cousins, 2013). As a result, there has been an increase in the concentration of farming units within the commercial agricultural sector with landholdings
8
Stellenbosch University http://scholar.sun.ac.za
being consolidated into larger farming units of ownership and production while smaller and less efficient farmers were forced out of farming (Vink, 2009). The average farm size of these commercial farms in South Africa was estimated to be 1349 hectares in 1996 and would have increased slightly over the past few years (Fenyes & Meyer, 2003). These farmers, who are mostly white, are able to compete globally in agricultural markets and earn income comparable to the highest income groups in the country (Pauw, 2007). The contribution of commercial agriculture to the national economy has decreased over time (Vink & Kirsten, 2003) with agriculture’s current share in GDP less than 3% since 2005 (Greyling, 2012).
The smallholder sector in South Africa is mostly found in the former homeland areas on very small landholdings (Groenewald & Nieuwoudt, 2003) and is found in a wide range of locations in South Africa from deep rural areas to townships, cities and on commercial farms (Lahiff & Cousins, 2005). Production tends to be mostly on a subsistence basis for household consumption (Aliber & Hart, 2009) from gardens, demarcated fields or open rangelands. Fenyes and Meyer (2003) suggest that the majority of these rural households consist of women, children and aged people. There is considerable variation in the sizes of land for smallholder cultivation but is generally extremely small and in the range of 0 - 1.5 hectares per household. Of these, a substantial proportion of households farm on less than 0.5 hectares and only a small percentage of households farm on plots larger than 5 hectares (Lahiff, 2000).
Smallholder farmers in South Africa face various challenges when it comes to agricultural production. Limited access to markets, production factors and credit combined with property right constraints and high transaction costs make life very difficult for these producers farming on small pieces of land (Ortmann & King, 2006). Smallholders are faced with a range of technical and institutional factors which influences access to markets. Smallholder farmers that do market their produce will mainly deliver to one of three destinations, namely fresh produce markets, informal markets and less frequently, supermarket chains (Baiphethi & Jacobs, 2009). 2.4 Defining Smallholder Farming Systems
The definition of smallholders varies between countries and agro-ecological zones and the notion of “small” varies in different contexts (Narayanan & Gulati, 2002; Dixon et al., 2003; Nagayets, 2005; Machingura, 2007). This explains the freqeunt and interchangeble usage of the term “smallholder” with “small-scale”, “subsistence”, “resource poor”, “small”, “low- income”, “low-input” and the list continues (Nagayets, 2005; Machingura, 2007). There
9
Stellenbosch University http://scholar.sun.ac.za
seems to be no universally accepted definition for smallholder farmers, but some definitions have been noted in the literature both globally and domestically.
2.4.1 Global Definition One of the more general approaches used to define smallholder farmers would be to assess the common characteristics of these farmers such as their land and capital access, exposure to risk and input technologies and market orientation (Chamberlin, 2008). For instance, Lipton’s (2005) definition of a smallholder is based on whether or not most of the labour is performed by members of the family, while Dixon, Abur & Watterbach (2003) suggest that the term smallholder relates to their limited resource endowment relative to the other farmers in the sector. The definition used by Ellis (1999) is incorporated with this which states that smallholder farmers are farm households that have access to land and rely primarily on family labour for production to produce for subsistence and/or market sale (Machingura, 2007).
The World Bank defines a smallholder as farmers with less than 2 hectares of cropland and those that have a low asset base (World Bank, 2003). Finally, according to Narayanan and Gulati (2002), a smallholder farmer is characterized as one who practises a mix of commercial and subsistence production and where the bulk of labour comes from the family. These definitions all have a similar theme and concentrate on the basic characteristics such as constraints to land and labour (Machingura, 2007).
The formal definitions listed above have often been used to define smallholder farmers around the world; however, another common approach to define smallholders is to use farm size (or livestock numbers). This is mostly justified by the availability of international empirical data that can be used for a more general classification of smallholders (Nagayets, 2005; Machingura, 2007; Thapa, 2009). Yet this definition does have limitations of its own. Important factors such as quality of resources, farming operations, managerial ability and disparities across regions are not controlled for in farm size. For example, a small farm that produces high value crops on irrigated land is not comparable to a small farm on marginal land that produces staple crops. Kirsten and Van Zyl (1998) go on to state that:
“Size is not a good criterion for defining small farms. For example, one hectare of irrigated peri-urban land, suitable for vegetable farming or herb gardening, has a higher profit potential than 500 hectares of low
10
Stellenbosch University http://scholar.sun.ac.za
quality land in the Karoo. Turnover, or rather the level of net farm income, determines the farm size category, not the land size.”
2.4.2 Smallholder Farming within South Africa There are various different general definitions for smallholder farmers in South Africa and the terminology used to refer to them has been inconsistent. Various authors have used descriptive words to classify smallholders and these terms have been used interchangeably (Ortmann & Machethe, 2003). The terminology used has often also been linked to the specific number of farmers in a specific group, which makes classification difficult. Furthermore, the lack of good quality data on smallholder farmers exacerbates this problem of smallholder definitions (Cousins, 2013).
The term “small-scale” is often used in South Africa to refer to black smallholder farmers characterised by non-productive, backwards, non-commercial and subsistence agriculture (Kirsten & van Zyl, 1998). It is often used as the broader term to refer to the total number of farmers or households involved in agricultural production on a relatively small scale. According to Coetzee (2003) South Africa has approximately 2.1 million small-scale farmers, while Vink (2009) reports that there were approximately 1.3 million smallholder households involved in farming. According to the National Department of Agriculture in 2001, this amounted to approximately 3 million farmers or individuals (NDA, 2001). The more general consensus, however, suggests that the total number of small-scale farming households in South Africa is approximately 2 million farming households (Aliber, 2005; Aliber & Hall, 2010; Aliber & Cousins, 2013).
This broad group of small-scale farmers or farming households can be sub-divided into two groups as indicated in Table 1. The National Department of Agriculture, Forestry and Fisheries have sub-divided these farmers in South Africa into two distinct categories in their 2012 Integrated Growth and Development Plan (IGDP) (DAFF, 2012b). Firstly, Emerging smallholder farmers are formally defined as those that are located in the former homeland areas and are predominantly black. This group consist of approximately 140 000 black farming households who are said to be more commercially inclined by marketing their produce (Aliber & Hall, 2010; Tihanyi & Robinson, 2011). Secondly, the IGDP suggests that the second group of farmers, known as subsistence producers, are defined as those approximately 2 million households involved in agriculture which only produce agricultural goods for own household consumption (DAFF, 2012b). These estimates seem to align closely
11
Stellenbosch University http://scholar.sun.ac.za
with the findings in the GHS of 2009 and 2010, which are given in Table 1. These are extrapolated estimates of the number of black farming households in 2009 and 2010 in the South African population using the sampling weights in the data.
Table 1: Number of black farming households in South Africa
2009 2010 Definition Number1 Share of Total Number1 Share of Total Subsistence Farming Households 2324379 94.59% 2307496 93.51% Emerging Farming Households 132843 5.41% 160043 6.49% Total 2457222 100% 2467539 100% Source: GHS 2009 and 2010 The latest estimate on households involved in agriculture comes from the 2011 census, reporting that 2.6 million black households were involved in farming in South Africa. The KwaZulu-Natal and Eastern Cape were the provinces with the most agricultural households with 24.9% and 20.7% of the total respectively (StatsSA, 2013). Smallholder agricultural activities are very much concentrated in the former homeland areas of South Africa and in rural areas. The results from the GHS 2010 and Census 2011 are considered to be the latest and most reliable indicators of the current number of smallholder farming households. Thus, the South African smallholder sector consists of between 2.5 and 2.6 million households, of which approximately 160 000 sell their produce to the market, while the rest produce for subsistence purposes. This is illustrated in Figure 1 below, giving the graphical representation of the terminology used to define smallholders in South Africa.
1 Population figures are weighted by sampling weights
12
Stellenbosch University http://scholar.sun.ac.za
Figure 1: Graphical representation of the terminology used to define smallholder farmers in South Africa
Source: Own compilation based on GHS, 2010; DAFF, 2012b; StatsSA, 2013 2.5 Smallholder livelihoods in South Africa
It is well known that smallholder farming systems in sub-Saharan Africa occur within an environment with diverse biophysical and socio-economic conditions. Rural households develop different livelihood strategies said to be conducive to the given opportunities and constraints in their specific environment (Tittonell et al., 2010). Ellis (1998) refers to this livelihood diversification among rural households as a process by which a diverse portfolio of activities and social support capabilities are chosen for improved living standards.
Rural livelihoods in South Africa can be characterised by an environment with diverse economic activities which could be farm or non-farm related (Alemu, 2012). It is clear that rural livelihoods are today impacted by both legacies of the past and more concurrent changes presently and include factors such as the ascendancy of supermarket retail and the increased social grant assistance in the recent past (Neves & Du Toit, 2013). Many rural people are either directly or indirectly involved in agriculture (Pauw, 2007) and their livelihood options would either be farm-related (livestock and crop production), off-farm (wage employment on
13
Stellenbosch University http://scholar.sun.ac.za
other farms) or non-farm (non-agricultural income sources such as wage employment and remittances) (Alemu, 2012). Thus, these households create a living from various sources such as production, labour, trading, and transfers. The latter is well known to form the backbone of South African rural families in the form of social grants and remittance payments (Perret et al., 2005).
Smallholder households, mostly in the former homelands, are typically poor, black (includes African, Coloured and Asian) agricultural households that struggle to support themselves with income earned from agricultural activities. Production is mostly aimed at providing staple foods for household consumption and can be produced on anything from gardens, demarcated fields or open rangelands. There is considerable variation in the sizes of land for smallholder cultivation and extremely small (Lahiff, 2000). According to Perret and Kirsten (2000) only 2.7% of the 70% who participates in agricultural systems relies on their farming activities for income. More recent estimates from the GHS seem to indicate that only 5.8% of all black farming households sell their produce (GHS, 2010). 2.6 Support to smallholder farming in South Africa
Support to the smallholder sector in South Africa was first introduced by the Development Bank of Southern Africa (DBSA) in the mid 1980’s (Aliber & Hall, 2010). This was done by providing Farmer Support Programmes (FSP) supporting smallholder farmers in the homeland areas (from 1987 to 1993) and provided support services in the form of inputs, capital, mechanisation, marketing, training and extension. These interventions, though not perfect, were formally ended when the homelands were reintegrated into the provinces of South Africa in 1994 (Aliber & Hall, 2010).
Since the formal ending of Apartheid, the policy objectives relating to smallholder farming in South Africa have been framed in a very broad perspective; to help smallholders to become commercial and to expand (Aliber & Hall, 2012). These efforts were aimed at rectifying the dualistic gap between the smallholder and large-scale sectors and to maximise the contribution of the agricultural sector towards growth and development in the economy (NDA, 2001). The Department of Agriculture came up with various policies to support smallholder farmers as this was said to give solutions to the longstanding problems of unemployment and rural neglect. Yet, these policy interventions have done very little to support this sector; some even contributed to its decline (Lahiff & Cousins, 2005). These
14
Stellenbosch University http://scholar.sun.ac.za
include the deregulation of commodity markets, failed land reform initiatives and the dismantling of development corporations.
In recent years, smallholder agriculture in South Africa has climbed the political agenda attracting new policy emphasis (Hall & Aliber, 2010). There could be various reasons for this, but Aliber and Hall (2012) suggest major food price inflation since 2000 and the need to secure food security within the country, were the main reasons. This message was very clearly expressed in the ANC Election Manifesto:
“The ANC government will: Intensify the land reform programme to ensure that more land is in the hands of the rural poor and will provide them with technical skills and financial resources to productively use the land to create sustainable livelihoods and decent work in rural areas. ... [And] expand [the] agrarian reform programme, which will focus on the systematic promotion of agricultural co-operatives throughout the value chain, including agro-processing in the agricultural areas. Government will develop support measures to ensure more access to markets and finance by small farmers, including fencing and irrigation systems. (ANC, 2009).
Statements like these have created a general rhetoric that smallholder farming will receive more direct support from government. When looking at the amount of public expenditure on the agricultural sector over the past 5 years, both at national and provincial level, it seems that increased support has materialized. Considering the fact that large-scale white farmers receive limited direct support from government, Aliber and Hall (2012) roughly estimate that this expenditure would amount to more than R2500 per ‘agriculturally active, black household’ using only provincial expenditure. However, in reality, the distribution of these government transfers rarely reach the hand of the farming households and are often highly skewed to favour certain farmers (Hall & Aliber, 2010). Table 2 follows the same procedure as the above mentioned paper, but shows agricultural support towards black farming households from 2009 to 2011 using the GHS survey.
15
Stellenbosch University http://scholar.sun.ac.za
Table 2: Government agricultural support to black farming households
2009 2010 2011 Share Share Share Type of support of of of Number2 Number2 Number2 Total Total Total (%) (%) (%) Training 49425 1.91 45058 1.56 74372 2.44 Extension visits 46477 1.79 59688 2.07 73998 2.43 Grants 5228 0.2 6347 0.22 10955 0.36 Loans 3379 0.13 6101 0.21 12103 0.4 Input as part of loan 7744 0.3 26706 0.93 26990 0.89 Inputs for free 52237 2.02 132394 4.6 148324 4.87 Livestock health services 261368 10.08 228811 7.94 204426 6.71 Other 1770 0.07 4801 0.17 14034 0.46 Total 336754 12.99 388898 13.5 368249 12.07 Source: GHS 2009, 2010 and 2011 From Table 2 it is clear that only approximately 13% of black farming households received direct support. This support came mostly in the form of livestock health services and secondly input supplies. The question that needs answering is how such substantial transfers from government have resulted in so little benefit to farmers. Aliber and Hall (2012) suggest a few explanations: Firstly, the lack of resources seems to be stretched when it comes to extension personnel who need to cover such a big proportion of farmers. Secondly, many black farmers are said to be almost invisible in the sense that many departments are unaware of their existence. Thirdly, there is a preference for quality instead of quantity and other related issues with under spending from government. 2.7 Conclusion
This chapter has given a broad overview of the South African smallholder sector. The first section explained the dualistic nature of the agricultural system and the various factors that have played a prominent role its development. This review indicates that distinct policy interventions, specifically the Natives’ Land Acts and the support towards white commercial farming, caused the duality within the sector. Prior to these measures, the African farming sector was competitive with the settler farmers and that these typically produced surplus food for the market. In the 21st century, even though these policies have been removed, the dualistic agrarian structure persists. African farming is characterised by lack of resources, relatively low production, and lack of access to markets.
2Population figures are weighted by sampling weights 16
Stellenbosch University http://scholar.sun.ac.za
The definitions used for smallholder farmers differ both internationally and domestically, with no universally excepted method to define them. Definitions typically relates to the characteristics of the farmer such as their access to land and capital, exposure to risk and input technologies, and market orientation. These tend to change in different contexts and various terminologies have used to simplify these definitions. In South Africa, smallholders are divided into “subsistence” and “emerging” farmers which relates to their reason for farming. The former consists of approximately 2.3 million households producing food for household consumption. The latter were said to sell their produce and this group of farming households were approximately 160 000.
South African rural households develop various livelihood strategies in accordance to their specific environment. Livelihood diversification takes place when individuals or households create a living from various sources such as production, wages, trading and transfers (grants and remittances) towards resilience and sustainability. Lastly, support towards smallholder farmers has been limited even though the National and Provincial budget spending on agriculture has increased. Challenges faced with support mechanisms relate to the limited extension personnel in relation to the number of farmers; the invisible nature of many of these farmers and the preference for quality instead of quantity.
17
Stellenbosch University http://scholar.sun.ac.za
Chapter
3 Literature Review on Typology Development
3.1 Classification within Agriculture
Classification is a process that is central to all facets of life (Bailey, 1994). In the simplest of terms, it is defined as the process of ordering entities into groups or classes based on similarities (Everitt et al., 2011). In particular it is the process of organizing complex and disparate data in order to provide a basis for analysis, decision making and reasoning (Xu & Wunsch, 2009; Everitt et al., 2011).
Agricultural systems are comprised of a basic production unit, i.e. the farm, which has its own distinctive limitations and constraints and faces a heterogeneous decision-making environment. Given this diversity within agricultural systems, various schemes of classification have been developed and evolved over time (van der Ploeg et al., 2009). The early schemes used structural characteristics, such as scale and factor intensity, in order to classify farming units. The rationale of these schemes was exclusively economic in nature with the main criterion between different farm types according to farm income (Andersen et al., 2007). For example, the European farm typology (1985) classified farms according to production orientation and economic size. To do this, the regional gross margin (standard farm income per production unit) for each type of agricultural production (whether crop or livestock) were assigned and multiplied by the volume of production to obtain the income from each production type on the farm unit. The proportion of each production type’s contribution to the gross margin was used to classify farmers (Andersen et al., 2007).
It was thought that farms would enter a modernization pathway that would merely be a quantitative process, with modernization occurring through an enlargement of farm sizes and a general increase in productivity, which would cause a convergence towards a new optimum. The outcome of this pathway was expected to cause increasing farm income over time as a result of structural development.
18
Stellenbosch University http://scholar.sun.ac.za
According to these early models it was expected that only the farmers adopting the newest technologies and realizing the highest levels of intensity and scale (thus higher incomes), would be able to stay in farming (labelled as the “series” farmers). It was expected that the “traditional”, “non-professional” and/or “small-scale” farmers would in time move out of farming as agriculture moved towards the modernization optimum. Initially, these classification schemes used characteristics such as regional location that were associated with varying historical trajectories, urban-rural relations, ecological conditions, landscapes and institutional structures. However, by the 1980’s it became apparent that the status quo classification system of farmers had two major deficiencies. First, differentiation occurs at the product-level despite similar structural characteristics. Secondly, a large amount of diversity among agricultural systems was not accounted for in these early models since smallholder and/or part-time farmers were largely ignored in the systematization of farms within Europe (Andersen et al., 2007; Van der Ploeg et al., 2009). Fundamentally, these classification schemes failed to deal with the multi-functionality of agriculture and as a result several deviations from these models contributed to the development of two new methodologies to classify agricultural systems (van der Ploeg et al., 2009). These include the Farming Styles approach, which was developed in the Netherlands; and the Farming Activity approach developed in France.
3.1.1 Farming Styles The basis of the Farming Styles approach is that heterogeneity in agriculture and amongst farmers is not random or just a consequence of physical characteristics or different structural factors affecting farmers. It is rather a reflection of social differences and the diversity can be explained by the manifestation of a range of farming styles (Howden et al., 2007). The approach was developed by Jan Douwe van der Ploeg (1990; 1993; 1994) at the University of Wageningen in the late 1980s and early 1990s. This theoretical approach essentially seeks to explain diversity amongst farmers in a particular setting (Howden & Vanclay, 2000; Howden et al., 2007). It is a way of conceptualizing farming as a social process within a specific cultural, economic, and political context as well as farm management components (Howden & Vanclay, 2000). A farming style is said to be a socially constructed type that reflects the worldview and strategy of one specific farming practice for a particular commodity in a particular region. Van der Ploeg (1994) elaborates:
19
Stellenbosch University http://scholar.sun.ac.za
“A Style of farming then is the complex but integrated set of notions, norms, knowledge element, experiences etc., held by a group of farmers in a specific region, that describes the way farming praxis should be carried out”.
Thus, categorization of farmers can be done by recognizing consistencies across the variety of different social components of farming. The use of farming styles theory in the development of farm typologies is well documented in the literature (van der Ploeg, 1994; Mesiti & Vanclay, 1997; Howden et al., 1998; Howden & Vanclay, 2000; Thompson, 2001; Vanclay et al., 2006; Howden et al., 2007).
The typology studies that have used the Farming Styles theory in their methodologies have emphasized the importance of the farmer as an individual and links land management decisions with social dimensions (Emtage, 2004). Farming Styles theory also places more emphasis on qualitative methods rather than the traditional quantitative techniques used to identify patterns and also often involves the farmer’s self-assessment developed by Whatmore (1994) through experiential, deductive reasoning (Howden et al., 2007).
3.1.2 Activity Systems In French agrarian sciences the agricultural systems approach to classification comes from a long lineage of farming systems research throughout the 1970’s and 1980’s. A farming system is defined as a population of distinct farm systems that have generally similar resource bases, enterprise patterns, livelihoods and constraints, and for which similar development strategies and interventions would be suitable (Kobrich et al., 2003; Madry et al., 2013). It is seen as consisting of a totality of consumption and production decisions by the farm household, including agricultural orientation, off-farm enterprises and consumption of food, and each one has its own unique farming system (Kobrich et al., 2003).
The farming systems concept was replaced by the “activities system” concept, which broadened the classification of farms by including other factors besides the basic agricultural activities of households (Cochet, 2012). The activity systems approach was introduced to account for the pluriactivity of farmers; the phenomenon where farming is done on a part- time basis and the farmer is involved in many other activities apart from food production (Bessant, 2006). Economists and agricultural economists found it difficult to find farming systems in Sub-Saharan African societies due to the embeddedness of the production process
20
Stellenbosch University http://scholar.sun.ac.za
with the household unit, the accumulation unit and the consumption unit of the household (Cochet, 2012).
The “activity” approach realised the need to include not only production functions, but also environmental and social functions to better assess the environmental and social dimensions of agriculture in rural development. Thus, in short, the rationale underlying farming systems can only be understood by making reference to the broader perspective, the activity system, which constitutes the sphere in which farmers’ practises and choices appear coherent (Cochet, 2012). This approach was accompanied by methodologies and concepts that (a) incorporated the multifunctionality of agricultural activities, (b) took on multidisciplinary approaches and (c) were based on broader geographical frameworks (van der Ploeg et al., 2009).
In the discipline of agriculture and rural sociology, an activity system is seen as a set of dynamic and structured interactions carried out by a social entity in a specific agro-ecological and social context (Gasselin et al., 2012). Under the activity classification system, the typologies not only highlighted variation in farm income sources; but also explained the motivations and aims of these activities and functions (van der Ploeg et al., 2009). The use of these classification systems was introduced from a development point of view where diversity was not considered an obstacle or constraint, but rather as an expression of the capacity of the agricultural system to adapt and sustain different scenarios (Laurent et al., 1999). It recognises that agricultural activity has an economic function which is not only driven by commercial farm income, and even if it is, can relate to different systems and norms. Furthermore, the activity system helps to explain the why and the how of a productive process in agriculture, even though agriculture might not be the primary activity of the household (Cochet, 2012). 3.2 Defining Farm Typologies
Classification schemes within agriculture have been widely used to describe and analyse diversity in agricultural enterprises (Emtage, 2004). It involves developing a set of formal categories into which a particular field of data is partitioned. In contrast, a typology is a particular type of rigorous classification in which a field of data is divided up into categories that are all defined according to the same set of criteria, and that are mutually exclusive.
In particular, according to Tefera et al., (2004) a typology is defined as a quantitative or qualitative procedure that categorizes households or individuals into homogenous groups,
21
Stellenbosch University http://scholar.sun.ac.za
which face similar constraints and incentives and are influenced by external factors in a similar way. In agricultural analysis, diversity within the rural environment manifests itself in various responses and the use of farm typologies is a useful way of describing this diversity. This is achieved by specifying the structural characteristics of different farm types, where each type or group is significantly different from the other in relation to a specified criterion (Laurent et al., 1999; Van der Ploeg et al., 2009). Therefore, the relevance of any farm typology will depend heavily on its ability to capture the diversity of farming systems through maximizing the homogeneity within groups and the heterogeneity between groups (Iraizoz et al., 2007). 3.3 Rationale for developing smallholder Typologies
In recent years many studies have focussed on defining farm typologies in various countries. (Dorward, 2002; Johnson, 2002; Kobrich et al., 2003; Emtage, 2004; Bidogeza et al., 2009). This is certainly the case with smallholder farming households in sub-Saharan Africa where production takes place in diverse socio-economic and biophysical environments (Tittonell et al., 2010). In this context rural farming households develop different livelihood strategies according to their different opportunities and constraints.
National governments in many countries are focusing on promoting sustainable natural resource management in terms of achieving objectives in environmental, social and economic development (Emtage, 2004). In this regard, typologies are widely used in the literature in order to understand structural changes in farming with regards to output, employment, farming intensity and impacts of policy reforms (Iraizoz et al., 2007). According to Landais (1998) typologies of farmers are constructed to help those who are administrating and designing development policies and programs in two ways. These are firstly to analyse the functionality of the farms and secondly, to provide useful recommendations on techno- economic matters which will help to optimise farming operations. It is generally accepted that the land management behaviour of farmers and rural households are not exclusively motivated by economic considerations such as maximizing the productivity of the farm unit (Emtage, 2004). 3.4 Theoretical Background for Creating Typologies
The theoretical understanding of the phenomenon of interest is used to determine the criteria that underpin a specific typology. This theoretical basis is crucial for defining the
22
Stellenbosch University http://scholar.sun.ac.za
relationships between the factors that influence the behaviour of households. According to Emtage (2004) there is a variety of theoretical perspectives that have been used to construct and develop typologies of farmers and rural households. These include Farming styles; Sustainable livelihood; Farming context and Market structure theory. All of these theories strive to account for the behaviour of individuals or households and each designates the behaviour as a consequence of the interaction between factors such as social, cultural, economic, institutional, biophysical and personal factors.
Farming styles theory Farming styles theory is already discussed earlier in this chapter; it relates to a distinct set of styles which farmers are acutely aware of and from which they make decisions. Studies that have used farming styles as a theoretical background emphasise the importance of the farmer as an individual in terms of decision-making, and tend to place more emphasis on qualitative rather than quantitative methods to identify different types (Emtage, 2004).
Farming context theory Farming context theory was used by Kaine and Lee (1994) who state that behaviour arises because of a combination of personal, social, biophysical and economic factors. This theory emphasizes examining differences in farming practices within the same type of agricultural enterprises and examines the evolution of the enterprise given the present resources, objectives and practises (Emtage, 2004).
Market structure theory The market structure theory has been used to create typologies of farmers and uses methodologies from marketing studies to guide the typology development. These seek to use typologies to analyse the diversity of consumers for a particular product. One such study includes Barr (1996) who tried to discover common features in the descriptions of farm types interviewed in the study, aiming to create a regional typology of defining the “market” for perennial pastures (Emtage, 2004).
Sustainable livelihood theory The Sustainable Livelihood (SL) approach to typology development has a rich history of over 50 years and has profoundly shaped rural development thinking and practice. It is multi- disciplinary in the sense that it incorporates insights from a wide range of disciplines including, political, sociological, agricultural, and/or environmental perspectives (Scoones,
23
Stellenbosch University http://scholar.sun.ac.za
2009). Thus, it includes complex interactions of how rural livelihoods intersect with political, economic and environmental processes.
Fundamentally, this approach is focused on the objectives, scope and priorities for development of low-income communities with the aim of identifying appropriate policy interventions to address their constraints and/or challenges (Farrington, 2001; Carney, 2003). As such the SL approach has been adopted in order to identify, design and assess new iniatives, to review existing activities, to inform strategic decision-making and for further research (Ashley & Carney, 1999). The SL approach incorporates three key elements. First, it is a set of principles that specify developmental activity which should be people-centred, locally differentiated according to relevant criteria and multi-level for the purpose of understanding livelihoods. Second, SL uses conventional analytical frameworks (economic, social, institutional etc.) that enable the identification of poor people’s options and constraints. Third, the developmental objective of SL should be clear i.e. to enhance the overall level of sustainability of livelihoods. In its application to agriculture, the SL approach has routinely been applied to the development of farming household typologies and is synonymous with farming systems analysis (Belsky, 1984; Perret & Kirsten, 2000; Dorward, 2002; Tefera et al., 2004; Perret et al., 2005; Babulo et al., 2008; Tittonell et al., 2010; Righia et al., 2011). The analysis focused on households rather than individual farms thereby recognizing the importance of the household as the primary decision-makers in livelihood choices. Thus, the household is seen as the decision-making hub and the outcome of the SL research is directed to improve the livelihoods of poor households. This is done by improving food security, cash income and the environment (Emtage, 2004). 3.5 Approaches to Constructing Typologies
There are three fundamental approaches used to construct typologies in the rural or farming context (Whatmore, 1994). These include; (1) taxonomic, a positivist approach that identifies typologies using empirical data; (2) relational, a realist approach which identifies groupings based on theoretical assumptions on structural relations; and (3) experiential, a hermeneutic approach using human reasoning to identify groups (Busck, 2002).
The taxonomic approach, also referred to as the ‘positivist’ approach, develops typologies on the basis of empirical data. This approach is used most frequently in developing rural typologies. Second, the ‘relational’ approach identifies groups by their coherent patterns of socio-economic relations by the object of study and its structural context in terms of
24
Stellenbosch University http://scholar.sun.ac.za
theoretical considerations. Third, the ‘experiential’ approach identifies groups by the interpretation of the human actors that inhabit the land to give meaning to certain ‘folk’ or ‘experiential’ groups (Emtage, 2004).
Perret and Kirsten (2000) give another perspective on the different types of typologies by distinguishing between a ‘structural’ and ‘functional’ typology. The ‘structural’ typology examines the factors of production and how these are structured, while the ‘functional’ typology relates to the decision making of farmers within their biophysical and social environment. The structural typology is equivalent to Whatmore’s (1994) relational approach, while the functional typologies are analogous to the taxonomic and experiential approach (Emtage, 2004).
More recently another way of classifying typology studies refers to the classification system used to develop the typology (Emtage, 2004). Previous typology development has followed one of, or a combination of, two main approaches found in the literature: the Qualitative Approach and Quantitative Approach (Iraizoz et al., 2007; Laoubi & Yamoa, 2009; Righia et al., 2011). How these relate to the approaches suggested by Whatmore (1994) and Perret and Kirsten (2000) will be explained in the next section.
3.5.1 Qualitative Approach Qualitative typologies are often based on a priori classification and depend on expert knowledge. These classification schemes, also referred to as deductive systems, rely on the knowledge and judgment of the researcher in order to define the specific segmentation of different groups according to their characteristics (Iraizoz et al., 2007). The focus of this approach is on identifying and describing what is typical for the different types of farmers instead of defining the boundaries that cause differentiation between groups (Van Averbeke & Mohamed, 2006). Studies that have applied the qualitative approach in the development of typologies include wealth rankings, farming styles as well as studies that created constructed types (Emtage, 2004).
Within the qualitative approach, typologies can be built on formal discussions (interviews) between researchers and those being researched in a participatory fashion. Those interviewed will then identify the important differences within the population to be used as criteria in the typology development (Emtage, 2004). Alternatively, in the qualitative approach, typologies can be developed by means of the researcher’s expert opinion to define types. These typologies are developed based on a priori knowledge by experts, followed by detailed on-
25
Stellenbosch University http://scholar.sun.ac.za
farm questionnaires, to develop a typology on the analysis of the patterns of responses in the quantitative data (Emtage, 2004). Both of these are said to be structural typologies according Perret and Kirsten (2000). In Whatmore’s (1994) classification the former corresponds to the ‘relational’ approach and the latter to the ‘experiential’ approach. Qualitative typologies have therefore most often been used in the farming styles literature.
The qualitative approach starts off with the establishment of the theoretical framework. After the theoretical framework has been identified, the next step in the typology development would be to select the criteria that will be used to measure differentiation between farm types. This is done by choosing the specific indicator variables that will be used in the analysis. The specific choice of variables will ultimately have the greatest influence on the results of the classification and is in itself a form of classification. The selected variables should be relevant and be investigated before being used in the classification scheme.
Once the theoretical framework and criteria have been selected, the researcher would then seek to formulate a provisional typology based on a priori classification that relies formally on the knowledge, understanding and judgment of the researcher to define the characteristics of the segmentation. These methods use mostly arbitrary and ad hoc considerations (Iraizoz et al., 2007; Gelasakis et al., 2012). Following the provisional typology by the researcher, interviews and surveys will follow on a number of the farms in the specific study area in order to verify each farm type and to establish whether or not the provisional typology is valid. Next, revisions of the provisional typology will be based on the results of the interviews until the researcher is satisfied with the results and will then produce a complete typology of the different types of farms. These are the common steps used in the qualitative approach and are illustrated in Figure 2 below.
One of the main advantages of using the qualitative approach lies in the actor-orientation towards the classification which makes sure that the farmers themselves can identify with the groups (Van Averbeke & Mohamed, 2006). Some of the disadvantages of this approach include a high dependence on the researcher; the inability to make full use of the available data; the lack of statistical foundation and the difficulty in reproducing these typologies (Iraizoz et al., 2007).
26
Stellenbosch University http://scholar.sun.ac.za
Figure 2: Steps used in both qualitative and quantitative typology development
Source: Own compilation based on Kobric et al., 2003; Emtage et al., 2005
3.5.2 Quantitative Approach The quantitative identification and characterization process utilize multivariate analysis and study diversity by using a finite number of variables to categorize farms, which is more precise and closer to reality (Van Averbeke & Mohamed, 2006). In recent years many studies have utilized the quantitative approach in order to create farm typologies (Ballas et al., 2003; Emtage et al., 2005; Bidogeza et al., 2009; Laoubi & Yamoa, 2009; Gelasakis et al., 2012).
The quantitative approach follows the same first steps as the qualitative approach as indicated in Figure 2. These steps resemble a combination of pathways used to create farming systems
27
Stellenbosch University http://scholar.sun.ac.za
types as proposed by Kobric et al. (2003) and Emtage et al. (2005). Figure 2 has been modified to also include the latest statistical techniques used to create typologies.
Step 1 and 2 in the quantitative approach, like the qualitative approach, involves the establishment of the theoretical framework and the variable selection. The selected variables should clearly relate to the features that the specific research is concerned with and the variables should have strong discriminating power which will lead to better classification of individuals (Emtage et al., 2005). It is important to note; across both qualitative and quantitative approaches, there is no universal rule for selecting specific variables since these would depend on the objective of the study. That being said, in farming systems classification the following variables are commonly used: farm size, income, capital, labour, resource management, production characteristics, soil quality and household demographics. According to Kostrowicki (1977) the best way to select the variables in a classification scheme is to use quantitative variables which are based on the choice of a limited number of variables of a specific characteristic and are universal, significant and representative of differentiation within these systems.
Step 3 involves the data collection process. This can be done by using either primary data or secondary data and would also be dependent on the specific research question. In previous studies secondary surveys that have been used include national household surveys or survey used in other studies (Dorward, 2002; Bidogeza et al., 2009; Takeshima & Edeh, 2013). Within this step, the data will also be checked so that variables with little variability, irrelevant to the specific typification, and/or show high levels of correlation should be eliminated as these would influence similarity measures between different groups (Kobrich et al., 2003). After the data is ready for analysis, the specific method to create the specific groups within the data is determined and applied in step 4. Consequently, the researcher can either move directly to Cluster Analysis (CA), or choose to use one of several data reduction tools or techniques. When CA is used directly after step 3, the data needs to be standardized by calculating the z-scores (Dorward, 2002; Jansen et al., 2003; Halder & Urey, 2003; Emtage & Suh 2005).
When CA is not directly applied to the data, several statistical methods have been used in step 4 as indicated in Figure 2. The most notable and frequently applied
28
Stellenbosch University http://scholar.sun.ac.za
methodologies include Principle Component Analysis (PCA), Multi-dimensional Scaling (MDS), Multiple-correspondence Analysis (MCA) and Categorical Principle Component Analysis (CatPCA) and Principle Component Analysis (PCA) (Dossa et al., 2011; Righia et al., 2011). These techniques are all used for data reduction purposes. Variables that are significantly correlated in the main datasets can be combined or transformed in order to generate fewer variables made up of synthetic factors (Ballas et al., 2003). The new dataset retains the important information in the raw data while also discarding much of the random statistical noise. The different techniques are mostly designed to be able to facilitate different data structures and formats that would best assist the specific research method. In the past, all of these methods have been used for quantitative farming typologies (Milan et al., 2006; Blazy et al., 2009; Dossa et al., 2011; Nainggolan et al., 2011; Righia et al., 2011; Gelasakis et al., 2012; Rouabhi et al., 2012).
Out of all of the abovementioned methods, PCA has been used most consistently in farming typologies worldwide (Machethe, 2004; Maseda et al., 2004; Bidogeza et al., 2009; Dossa et al., 2011; Nainggolan et al., 2011; Madry et al., 2013). These techniques are all used to define the underlying structure in the data matrix and analyse the nature of the linkages between the large variables by a number of dimensions called factors (Iraizoz et al., 2007).
In Step 5 CA is applied to either the original standardized data or the new data factors created in Step 4. Cluster Analysis refers to a set of multivariate techniques that seek to classify objects (individuals, households, products etc.) according to their characteristics into groups (Hair et al., 1998). The literature on CA and its uses are both voluminous and diverse as this technique has been used by almost every fields of study. The terminology of cluster analysis even differs in the different field of study. Biologists and researchers in the natural sciences would often refer to it as “numerical taxonomy”, while sociologists and economists mostly refer to it as “typologies” (Anderberg, 1973). According to Makhura et al. (1999), CA is better known to be exploratory rather than a hypothesis-testing tool and is used to create groups based on measures of closeness. When the specific similarity measurement has been selected, the researcher will then decide between the two main algorithms used for clustering; Aggregative Hierarchical Clustering (AHC) and Non-Hierarchical Clustering (NHC), which are not mutually exclusive. Often these two methods have been used together
29
Stellenbosch University http://scholar.sun.ac.za
which allows for one to benefit from the advantages associated with both while also minimising the drawbacks of each separately (Iraizoz et al., 2007).
The final step in the Quantitative typification comes in the form of a validation of the results from the CA. It is important that these groups are stable and not merely imposed on the data by the classification process (Kobrich et al., 2003). There is no formal method to test the significance of optimality of the groups. Yet, one alternative is to contrast the groups according to the original hypothesis about its specific structure and with the researcher’s perception of the observed empirical results. However, another method that has been often used as a means of cluster validation is one-way analysis of variance (ANOVA) tests. This enables the researcher to identify differences in variances between the clusters from the variables used in the classification (Maseda et al., 2004; Gaspar et al., 2008; Bidogeza et al., 2009; Gelasakis et al., 2012).
For the purposes of this study, a quantitative, inductive typology will be developed by using Whatmore’s (1994) positivist approach. PCA is selected due to its consistent application in farming household typologies and because reduction in the dataset is needed. Furthermore, applying PCA in this study transforms the dataset into a new data matrix that does not require standardization of the variables and can be directly used in the next step of the analysis (Dorward, 2002). Hierarchical and non-Hierarchical CA will be used for the typology development, which allows for the benefit from the advantages of both and at the same time minimize the drawback associated with these individually (Punj & Steward, 1983). Lastly, CA is selected in this study as this technique has been successfully applied to typology development of farming households, in South Africa and other countries. 3.6 Conclusion
This chapter gives a literature review on the development of classification systems. From section 3.1, it is clear that the process of classifying objects into similar entities is central to life and is used for ordering complex data. Within agricultural systems, various schemes of classification have been developed over time to account for diversity. The earliest models wanted to classify farms according to its production orientation and economic size, expecting farms to develop on a modernization pathway towards larger and more productive farms (Andersen et al., 2007). One of the fundamental deficiencies of these classification systems were the lack of dealing with the multi-functionality within agriculture, with gave rise to new
30
Stellenbosch University http://scholar.sun.ac.za
methodologies to classify agricultural systems such as the Farming Styles and Farming Activity approaches (Van der Ploeg et al., 2009).
According to Tefera, et al. (2004), a typology is defined as a quantitative or qualitative procedure that categorises farmers into homogenous groups, based on a certain criterion. The rationale for creating farmer typologies is to better understand structural changes in farming concerning output, employment, farming intensity and the impacts of policy. Furthermore, typologies can improve rural development planning by firstly analysing the functionality of a farm, and to provide recommendations on techno-economic matters (Emtage, 2004). From the literature, various theoretical approaches have been used in the development of farming typologies as summarized in section 3.4. These are the Farming styles, Farming context, Market structure and Sustainable Livelihoods (SL) theories. Of these, the SL theory has a history of more than 50 years and is a multi-disciplinary approach that includes complex interaction of how rural livelihoods intersect with political economic and environmental processes. It focusses on the objectives, scope and priorities for development in low-income communities with the aim of identifying policy interventions to address constraints and/or challenges in rural areas (Carney, 2003).
Section 3.5 investigated the approaches used to create typologies, which can be divided into two distinct groups: qualitative and quantitative approaches. Qualitative approaches are said to be deductive classification systems and responds to patterns in qualitative data. Whatmore’s (1994) ‘relational’ and ‘experiential’ falls in the qualitative approach as well as Perret and Kirsten’s (2000) structural typologies. Quantitative classification systems are more complex and utilize multivariate analysis in order to create typologies. Six steps have been identified which were summarized in section 3.5.2. The first two steps entail the establishment of the theoretical framework and the variable selection, while various methodologies can be used to create typologies in the fourth step. PCA, MCA, MDS were all data reduction techniques used to reduce the dimensions in the data. Various studies have not used step 4 and applied CA directly from the original data (Dorward, 2002). CA techniques can be divided into Hierarchical and non-Hierarchical clustering procedures, with a number of typologies using both of these in the same study to minimise the drawbacks from each of these individually. CA will then yield a cluster solution of farmers or farm households based on the selected criterion. To establish the validity of these groups, ANOVA testing is often applied to test whether or not these groups are statistically different from one another in terms of the selected variables.
31
Stellenbosch University http://scholar.sun.ac.za
This chapter therefore concludes that a quantitative classification, applying PCA and CA would be best suited for the development of a typology of smallholder farming households in South Africa. This is due to data considerations, the need for dimensional reduction and the fact that these techniques have been successfully applied to smallholder typologies to assess livelihood strategies for rural farm households.
32
Stellenbosch University http://scholar.sun.ac.za
Chapter
4 Methodology
4.1 Introduction
This chapter concentrates on describing the methodology used to create the proposed household typologies for South African smallholder farms situated in the former homeland areas. After careful inspection of the different methods used to create typologies and considering the objective of the research, the quantitative approach towards typologies was selected. As indicated in Chapter three, several different multivariate methods can be applied to create farming typologies. In order to segment smallholder farming households Principle Component Analysis (PCA) and Cluster Analysis (CA) have been selected as the appropriate method to create groups. 4.2 Methods
4.2.1 Step 1: Theoretical Framework The theoretical framework adopted in this study is the Sustainable Livelihoods (SL) approach introduced in Chapter 3. The SL approach is a multi-disciplinary approach which seeks to not only look at agriculture, but includes economic, social, environmental and political perspectives (Scoones, 2009). According to Ashley and Carney (1999) SL approaches are based on the ever-changing thinking about poverty reduction, the way poor people live their lives and the importance of structural and institutional issues. The premise of sustainable livelihoods is that the effectiveness of development undertakings can be improved by 1) systematic analysis on poverty and its causes; 2) a better informed understanding of the opportunities for development and its impact on livelihoods and 3) placing people and the priorities they define as the central part of the analysis (Ashley & Carney, 1999).
This framework is chosen due to its strong association with rural development research and its strength in describing diversity at a community level. The SL approach is focussed on the objectives, scope and priorities for development in low income communities and is typically aimed at a household level rather than individual level (Emtage, 2004). Rural areas are commonly characterised by the presence of diverse economic activities, whether related to
33
Stellenbosch University http://scholar.sun.ac.za
agriculture or not (Alemu, 2012). It is a well understood that rural people in Sub-Saharan Africa have a tendency to move away from natural resource-based occupations (Ellis, 1998) towards non-farm activities. This phenomenon is also found in rural South Africa where this phenomenon leads to diversified rural livelihood systems (Perret et al., 2005). Even though the majority of rural households produce agricultural products, only a very small percentage of these rely on agriculture as a main source of income. Thus, livelihood diversification is a process by which rural families construct a diverse portfolio of activities and social support capabilities in order to secure more sustainable and resilient livelihoods (Ellis, 1998).
This study focusses on socio-economic features at household level of farm households in the rural former homeland areas of South Africa. In order to understand diversification among farming systems in these areas, SL allows for dynamic analysis of the different strategies farming households undertake to attain a higher standard of living. This study uses this multi- disciplinary framework by looking at various characteristics that would define specific farm systems and includes income and expenditure, household characteristics, production activities and food security measures. SL recognises that a specific livelihood encompass more than just income (cash and in kind), but includes social institutions (kin, family, village etc.), gender relations, property rights and a few others which would influence the strategies adopted by rural households (Ellis, 1998).
This approach recognises the importance of the household as the decision-making unit which base decisions on the households’ available resources, objectives, personal and socio- economic views and the rules and norms of institutions that govern the use of resources available to the household (Emtage, 2004). This framework has a been used extensively in the development of farm typologies (Belsky, 1984; Perret & Kirsten, 2000; Dorward, 2002 Perret et al., 2005) in the past and the outcome of the SL approach is designed to improve the livelihoods of poor households by improving their levels of well-being, food security, income and biophysical environment (Emtage, 2004).
4.2.2 Step 2: Data and study area In terms of small-scale agriculture, there exists little, detailed national-level data. To date, the census of agriculture, conducted by Statistics South Africa, only includes formal, commercial farms in the sample set. Datasets that does include smallholder farming on a national level includes the Labour Force Survey (LFS), National Income Dynamics Study (NIDS), Rural Survey, General Household Survey (GHS), Income and Expenditure (IES) and Census 2011
34
Stellenbosch University http://scholar.sun.ac.za
(Aliber, 2009; May & Carter, 2009; StatsSA, 2011a; StatsSA, 2013). Of these, the GHS and IES have more recently included detailed information about farming households. Therefore, in order to develop two distinct typologies of farming households in the former homeland areas of South Africa, the 2010 General Household Survey (GHS) and the 2010/2011 Income and Expenditure Survey (IES) were utilized in this study.
The GHS is an official nationally representative household survey administered by Statistics South Africa (StatsSA) which has been conducted annually since 2002 (StatsSA, 2011a). The latest publication for the purposes of this analysis was for 2010; which consisted of 6448 households that were agriculturally active, household, and will be the one used for the study. Since 2009, the GHS has included more detailed questions on smallholder household production in South Africa (StatsSA, 2011b). Examples include mostly household questions such as: what kinds of products were produced; did the household sell produce; did the household receive government agricultural support; size of land; stock numbers; where does the household plant crops and on what basis does the household have access to land.
The IES is also a national household survey conducted by StatsSA but is only conducted once every five years. This survey captures income and expenditure data at the household level, which is then used to calculate the Consumer Price Index (CPI) of goods and services in South Africa (StatsSA, 2012). This dataset also includes detailed questions regarding smallholder agriculture, not only on their income and expenditure on household goods, but also on their agricultural production systems.
The sampling framework for both of these surveys is based on StatsSA’s Master Sample (MS) that is based on the Population Census of 2001 enumeration areas. Both of these surveys had a two-stage stratified design sample with a probability proportional to size selection of the primary sampling units (PSU) in the first stage. After allocating the sample into provinces, the primary stratification was defined by geographic area type, which was divided into metropolitan and non-metropolitan areas. Secondary stratification was based on the 2001 Census data and used the following variables: household size, education, occupation, gender, industry and income (StatsSA, 2011b).
In order to sample smallholder farming households located in the former homeland areas of South Africa, Geographic Information Systems (GIS) were used and applied to both the GHS and IES. A geographic shape file from the Department of Rural Development and Land
35
Stellenbosch University http://scholar.sun.ac.za
Reform (2004) was used to locate the former TBVC/Homeland areas as they were spatially administered under the Land Acts of 1913 and 1936 (Table 3).
Table 3: Land area of the Former Homelands
Former Homeland States Hectares % of Total RSA Transkei 4426338 3.63 Bophuthatswana 3801642 3.12 KwaZulu 3606063 2.96 Lebowa 2217131 1.82 Ciskei 799223 0.66 Gazankulu 739838 0.61 Venda 648729 0.53 Kangwane 351214 0.29 Kwandebele 325893 0.27 Qwaqwa 104985 0.09 Total Area 17021056 13.96 Source: DRDLR, 2004
The former Homeland/Bantustan areas consisted of 10 distinct “states” which made up 13.96% of the total 122.1 million hectares of land in South Africa. Out of the ten former “states”, the Transkei area was the largest with 4.42 million hectares, followed by Bophuthatswana and KwaZulu with 3.80 and 3.61 million hectares respectively. The geographic boundary information of the former homeland areas allows for the effective sampling of households located in the former homeland areas. This was done by using the smallest geographical unit, Enumerator Areas (EA) that enumerate or divide a country for census purposes, which are used in StatsSA’s survey methodology to spatially administer these surveys (Mokgokolo, 2011). Thus, each household would have an EA number and can therefore be broadly spatially located.
Using GIS techniques, all of the included files were transformed to the Hartebeesthoek 1994 Geographic Coordinate System (GCS) for consistency and the “joins” and “relate” function in ArcGIS (geographic computer software) were used to join the relevant geographic information together. In other words, by joining the EA codes together with the boundaries of the former homeland areas, only EA’s whose centroids (most central point) were located inside the former homeland boundary were selected. This enabled the effective extraction of only the EA’s that are situated inside these invisible boundaries. Using this new “Attributes Table” in ArcGIS, it is possible to extract these EA numbers as a variable into the main survey datasets and to generate a dummy variable (value of 1 if inside and value of 0 if not)
36
Stellenbosch University http://scholar.sun.ac.za
for every household within the GHS and IES respectively. The result of this process is given in Map 1 below. Here the former homeland areas are indicated in dark grey while all the EA’s within South Africa are indicated with light grey lines. The small, darkened triangles represent the specific EA’s that were sampled within the GHS, while the white dots indicated those from the IES.
Map 1: Location of homelands and enumerator areas in GHS 2010 and IES 2010/11
Source: Own compilation based on DRDLR, 2004
The sample used in the study thus included only households situated geographically within these areas. Further, for both the GHS and the IES datasets, only households involved in agricultural production within the specific survey year and which were located in the former
37
Stellenbosch University http://scholar.sun.ac.za
homeland areas were identified and extracted to form the sample set used in the analysis (see Table 4 below).
Table 4: Sample size of smallholder households
Survey Number of households Population GHS 2010 3540 1585683 IES 2010/2011 2999 1403411 Source: GHS 2010 and IES 2010/11 It is important to note is that when these households were identified, 99% were listed as living in black (household head being black) households in both of these surveys. It was decided to therefore only include the black farming households in this study as this was the majority, in the context of separate development and the South African history of the study area. The data was checked for both samples and missing values for the relevant variables and data corrections reduced the sample for the analysis to 3540 households in the GHS sample, while the IES sample had 2999 households as indicated in Table 4. By using the household weights this amounted to approximately 1.59 million households in the GHS and 1.4 million households in the IES sample.
4.2.3 Step 3: Variable selection The selection of the variables is one of the most important steps in the analysis and will have the greatest impact on the ultimate results (Kobrich et al., 2003). Based on the selected SL framework, variables were selected to describe differentiation between households in terms of livelihood strategies. According to Kobrich et al. (2003) variables selection is generally based on three grounds; 1) the researchers’ experience and knowledge of the study area, 2) the objectives of the specific typology and 3) the quantitative information that is available. Variables selection in this study is also guided by the variables selection in various other farm typologies found in the literature. Important to note; the data from both the GHS and IES had only a limited number of variables that could be used in the analysis as these survey instruments are not designed to specifically facilitate typology development and needed to be appropriate for the multivariate statistical analysis.
Prior to PCA, the two datasets used in the analysis were investigated for the appropriateness of PCA; to ensure that variables were dependent on each other and that they were not too strongly correlated to one another. Two tests generally applied in the literature to assess the validity of the data and variable selection for PCA were used (Bidogeza et al., 2009). The Kaiser-Meyer-Olkim (KMO) (1970) measure for sampling adequacy was firstly used which
38
Stellenbosch University http://scholar.sun.ac.za
measure the strength of connection between variables and gives a value between 0 and 1. This measure is a summary that indicates the proportion of variance in the selected variables that might be caused by underlying factors. A value of 0 indicates that the sum of partial correlations is large relative to the sum of the correlations, suggesting that a form of factoring the selected variables would be inappropriate. A value close to 1 would however indicate the correlation between variables is relatively compact and PCA should yield stable and reliable factors (Field, 2009). The general rule states that if the KMO value is greater than 0.5, the amount of correlation among the selected variables in the data is acceptable. Thus, a KMO measure of greater than 0.5 would suggest that the selected variables are indeed dependent on one another and acceptable for use in PCA.
Secondly, Bartlett’s test of sphericity was used which tests the null hypothesis that the original correlation matrix of the proposed dataset is an identity matrix. If this is the case, all the correlation coefficients between the variables would be zero and it shows that all the variables are perfectly independent from one another and therefore does not add to the proposed finding of factors that measure similar things (Field, 2009). Thus, a p-value of smaller than 0.001 would suggest that the null hypothesis is rejected, indicating that the correlation matrix is significantly different from an identity matrix, and that there is sufficient correlation between variables (Iraizoz et al., 2007). Both of these tests were performed and applied to the two datasets used in the study. These were applied based on the fact that these two tests are standard statistical test applied to PCA analyses and for its wide application in farm typology studies that have used PCA in its classification systems (Maseda et al., 2004; Bidogeza et al., 2009; Riveiro et al., 2013)
These selected variables are listed in Table 5 which aggregates these variables into four categories; namely, household characteristics, income & expenditure, production characteristics and food security. This Table also shows the different variables used in the development of the typology from both the GHS and IES datasets.
39
Stellenbosch University http://scholar.sun.ac.za
Table 5: Variables used in the typology of smallholder farming households
Category of Variable Name Definition GHS IES Variable
Head_age age of head Yes Yes Head_education years of education of head Yes Yes Head_gender gender of head Yes Yes Married marriage status of head Yes Yes characteristics HHsize size of the household Yes Yes Grant_receivers number of grant receivers in hh Yes No Economically_active number of economically active in hh Yes No
Household HH_children number of children in household Yes Yes
Income_salary monthly total hh salary income Yes Yes
and Income_childgrant monthly total hh child-grant income Yes Yes
Income_oldagegrant monthly total hh old-age grant income Yes Yes Income_remittance monthly total hh remittance income Yes Yes Income Expenditure Expenditure_total monthly total hh expenditure Yes Yes Hectares size of farming land Yes No Land_farmland produced on farmland Yes No Land_backyard produced in backyard Yes No Inventory_cattle cattle livestock units Yes No
Inventory_sheep sheep livestock units Yes No Inventory_goats goat livestock units Yes No Inventory_chicken chicken livestock units Yes No Orientation Agric_gov_support did household receive farmer support Yes No Whyprod_mainincome why produce, as main income source Yes No Whyprod_extraincome why produce, as extra income source Yes No Whyprod_extrafood why produce, as extra food source Yes No Production Business_act business activity No Yes Inputcost_crop monthly crop input expenditure No Yes Inputcost_livestock monthly livestock input expenditure No Yes Inputcost_services monthly agricultural services expenditure No Yes
Hunger_child+adult hh members went hungry during year Yes No y
Food Percap_food_exp total food expenditure divided by hh size No Yes securit
Household characteristics In livelihood analysis, the household demographics play an important part in understanding diversity and have routinely been included in typology research (Perret et al., 2005; Bidogeza et al., 2009). The variables included in this category are the age, education, gender and marital status of the household head as well the household size, the number of grant receivers and economically active households members, these matter because a livelihood
40
Stellenbosch University http://scholar.sun.ac.za
encompasses social and kinship networks for facilitating diverse income portfolios as well as gender relations (Ellis, 1998). Furthermore, the household characteristics would also influence the different income sources within the household and would therefore influence livelihood decisions.
Income and Expenditure The income and expenditure variables are very important for the analysis. In this study, only the main income sources of the rural households were included. These were salary income, old age grant income, child grant income and remittance income and included as monthly rand values. Expenditure was included in the study as total monthly household expenditure, because this was the only question asked regarding expenditure patterns in the GHS. For consistency purposes, this exact variable is also used even though more expenditure variables are available in the IES survey.
Production Orientation These variables typically characterise the production systems within households. In this category there are big differences in measurement between GHS and IES. Again, the selection of variables is dependent on the specific dataset and the questions asked during the survey year. In the GHS, farm characteristics include farm size; number of cattle, sheep, goats and chickens; the place where crop production takes place (farmland or backyard) and the reasons for production as binary variables. The other two included variables relates to whether or not households received agricultural support from government and whether or not a household was involved in business activities.
In the IES dataset, production orientation is defined by the expenditure patterns on farming. This would give valuable insight into the farm cost structures of the household and is divided into crop, livestock and services inputs. These were the only variables that characterised households in the IES on their production orientation and are included as total monthly household expenditure on these inputs.
Food Security The most basic definition of food security refers to the ability of an individual to obtain or have access to sufficient food (Du Toit, 2011). To determine food security has, however, become a complex exercise because of the multiple definitions and indicator that exist in a wide range of disciplines (Altman et al., 2009). In South Africa, various different methods to access food security at a household level have been used (De Cock et al., 2013). Of these, the
41
Stellenbosch University http://scholar.sun.ac.za
GHS and IES are often used to assess household food security and each will probe a different dimension, as is the case in the selection of indicators in this study (Aliber, 2009; Altman, et al., 2009). In the GHS, food security is based on the self-reported experience of food security by the prevalence of hunger within households. The variable included in the GHS is binary giving a value of 1 if the household had an adult or child who indicated that they experienced hunger in the last 12 months3.
In the IES, a variable was created and includes the per capita expenditure on food within each household. It is expected that this variable would indicate the levels of food security in the sense that food insecure households are expected to spend less money on food per individual within the household. The complex and multi-dimensionality of food security makes it very difficult to measure and no perfect singe measure exist that capture all aspects of food insecurity (Webb et al., 2006). However, these variables are still important and included as indicators to account for food insecurity in the development of the household typology.
4.2.4 Step 4: Principle Component Analysis The central idea behind Principle Component Analysis (PCA) involves the reduction of dimensions found in a set of data containing a large number of interrelated variables, while simultaneously retaining the maximum amount of variation in the dataset (Jolliffe, 2005). This is accomplished by transforming the data into a new dataset comprised of a new set of variables, the principle components, which are scores, calculated for the underlying dimensions in the data. These resulting components are then syntheses of the original raw data and by using these new variables will avoid the need to standardise or transform the variables for the next step in the analysis (Gaspar et al., 2008).
To generate these synthesized data sets, a statistical technique, which condenses the selection of initial set of variables into a smaller number of discrete, non-correlated components or set of factors, is undertaken (Jackson, 1991; Everitt & Dunn, 2001; Joliffe, 2005; Nainggolan et al., 2011). The resulting absence of correlation between the factors is a useful property indicating different dimensions in the data (Manly, 1986).
The analysis starts by taking p variables X1, X2,...,Xp, across n-households and finding
combinations of these to produce a new set of indices, Z1, Z2,…, Zn, that are uncorrelated
3 Hunger is defined as “Always” or “Often” going hungry due to a lack of sufficient food in the preceding 12 months. The variable of interest presented here considers the question for adults and children, though they are similar when analyzing, and are also robust to classifying “Sometimes” as hungry.
42
Stellenbosch University http://scholar.sun.ac.za
(Manly, 1986). The first principle component is then the linear combination of the variables
X1, X2,…., Xp, and is given by: